Thermal Short-Circuit Testing for Large Cross-Sectional
Area Components in Modern Power Systems
P. Krasselt1, M. Bächle1, R. Badent1 and T. Leibfried1
1
Institute of Electric Energy Systems and High-Voltage Technology
Karlsruhe Institute of Technology
Engesserstr. 11, 76131 Karlsruhe (Germany)
phone:+49 721 608 43155, e-mail: Peter.Krasselt@kit.edu, Rainer.Badent@kit.edu, Thomas.Leibfried@kit.edu
Abstract.
TABLE I. - Wind power stations: network voltage
The integration of decentralized renewable
energies sources in the electric power system has an impact on
electrical components used in transmission and distribution
networks. In general, components with high ampacity and large
cross-sectional areas are introduced for network connection of
large renewable power plants. These components require
electrical type testing according to IEC standards. In this paper,
a real-time control method for thermal short-circuit testing is
proposed. It features automated short-circuit current duration
regulation to meet conductor target temperature, which is robust
against the phenomena of nonlinear resistance increase due to
conductor heating and network voltage sags. Inrush currents are
reduced using a phase-lock loop for optimal switch-on time
synchronization. For user convenience, the solution features
additional automated reporting. Experimental test data on a test
stand with an ampacity up to 40 kA validates the concept.
NETWORK VOLTAGE
Extra high voltage network (EHV)
Network coupler (EHV-HV)
High voltage network (HV)
Network transformer (HV-MV)
Medium voltage network (MV)
Distribution transformer (MV-LV)
Low voltage network (LV)
CAPACITY
1.35 GW
0.16 GW
10.95 GW
5.03 GW
15.85 GW
0.08 GW
0.04 GW
SHARE
4.0 %
0.5 %
32.7 %
15.0 %
47.4 %
0.3 %
0.1 %
2. Thermal short-circuit testing
Electric components are subjected to a variety of
electrical, mechanical and thermal stresses. Complex test
sequences are defined in product standards to replicate
these stresses, and new component designs are type
tested according to the relevant standards to ensure safe
operation and a long operational life.
Thermal stresses are caused by the Joule effect given in
(1): due to the flow of an electric current I for the time t
through the conductor with resistance R, a proportional
amount of heat Q is generated.
Key words:
Power system faults, thermal short-circuit
currents, wind energy integration, high-current testing, IEC
standards.
1. Medium voltage components:
Trend to large cross-sectional areas
Over the last 15 years, a renewable production capacity
exceeding 90 GW (2014) has been installed in Germany.
Outnumbering the connected thermal power plant
capacity of 94 GW (2013) in the next years, the
installation of decentralized renewable electricity
generation also defines new requirements concerning
components for electrical transmission and distribution
networks.
To illustrate this point, we will have a closer look on
network connections of wind power stations. The
publicly available information of installed renewable
electricity generation [1] funded under the German
Renewable Energy Act is evaluated for the network
connection level of each wind turbine generator system.
The evaluation shown in TABLE I reveals that nearly
half of the installed wind energy capacity is connected to
medium voltage (MV) networks.
In the time before the massive integration of renewable
electricity generation, standardized conductor sizes up to
a cross-sectional area of 300 mm² were sufficient for
nearly all challenges in MV network planning.
Nowadays, wind farm network connections require a
high ampacity, demanding MV products with crosssectional areas of 500 mm² and more [2,3].
Q  I 2  R t
(1)
During undisturbed network operation, conductor
temperature varies slowly depending on its actual current
and ambient conditions. In contrast, conductor
temperature rises extremely fast during short-circuits
until the fault current is interrupted by power system
protection. An adiabatic heating without heat dissipation
to the environment can be assumed, reaching conductor
temperatures up to 250 °C during the fault current
interruption time, which is in the range up to several
seconds.
Testing the thermal short-circuit withstand capability of
new designed components is therefore an important test
procedure. All components directly exposed to the shortcircuit current are subjected to this test, e.g. connectors
and lugs [4]. Thermal short-circuit testing is also
obligatory for accessories (e.g. terminations, straight or
branch joints) [5], as their insulation system is stressed by
high conductor temperatures after short-circuits.
158
EEEJ, Vol.1, No. 3, May 2016
The method for thermal short-circuit testing is
standardized [6]. The test objects are at ambient
temperature and are then heated to maximum permissible
temperature during short-circuit. For cross-linked
polyethylene (XLPE) insulation systems, which dominate
the MV market, a target temperature of 250 °C has to be
reached within a maximum time of 5 s.
Special attention is required to attain the target
temperature as accurately as possible. A direct
measurement of TOs temperature using thermocouple
elements is very difficult and uncertain due to electromagnetic disturbance emitted by the short-circuit current.
Therefore, the amount of heat to reach target temperature
is theoretically calculated [7], assuming a linear increase
of resistance due to heating.
B. Proposed closed loop approach
Testing products with increasing cross-sectional areas
requires more powerful test equipment for thermal testing
as the test current varies more nonlinearly.
Computational power and analog-digital-converter
performance currently available enable a regulation of
the required heating time in real-time by an online
evaluation of the short-circuit currents.
In this paper, a new real-time control concept for thermal
short-circuit withstand tests according to the test methods
defined in [6] is introduced. To achieve the required
temperature of the test object with high accuracy, a
closed loop control is proposed, using a current
measurement as a feedback variable.
For optimized operation of the high-current test circuit
and minimized MV network disturbance, the elements of
the high-current test setup are analyzed in the following.
3. Synthetic high-current test circuit
The synthetic high-current test circuit consists of the test
object, which is connected to a high-current transformer.
To control the 2 MVA high-current transformer, a 20 kV
circuit breaker on its MV side is used. For evaluation, the
current in the test object is measured by a current probe.
The structure of the test setup is shown in Fig. 1.
C. Properties of the high-current test setup
1) High-current transformer. For test objects with
high cross-sectional area, the impedance is
nearly purely inductive. To minimize inrush
currents, a phase synchronous switch-on
command at the voltage minimum or maximum
is preferable to reduce inrush currents of the
connected circuit. The transient inrush current of
this magnitude will cause a short network
voltage sag on the MV side. Furthermore the
remanent magnetization of the transformer core
has to be considered. Fig. 2 shows the hysteresis
diagram of a typical transformer core.
When a remanence has been build-up as a sideeffect by the previous test cycle, a current of
same polarity as the remanence may lead to
saturation effects and an excessively high inrush
current, too. As it is practically impossible to
measure the residual magnetism, the control
system has to estimate it using the recorded
waveform of the electrical current of the
previous test cycle.
2) Test Object (TO). Usually multiple test loops of
the TO are subject to testing. As shown in
Fig. 1, the TO is modelled by a resistance and an
inductance. The resistance is determined by the
conductor material, in most cases copper or
aluminum, and its temperature.
Fig. 1. High-current test setup
A. Conventional approach: open loop time control
In the conventional approach, the initial current is
measured with a short-time short-circuit. Assuming a
linear increase of resistance due to heating, the full time
to reach the target temperature is calculated. Once the test
object reached ambient temperature again, a timecontrolled short-circuit is applied to the test object with
the calculated duration.
This method suffers from the drawback that network
voltage variation and voltage sags during high-current
testing are neglected. In addition, the short-circuit current
amplitude decreases due to resistance increase with
higher conductor temperature (TABLE II). This effect
reduces networks voltage sag, resulting in a nonlinear
dependency between current amplitude and conductor
temperature, which degrades the accuracy of the open
loop time control.
TABLE II. - Typical increase in resistance
MATERIAL CONDUCTIVITY / (Ω/m)
20 °C
250 °C
Aluminium
2.82×10−8
5.44×10−8
Copper
1.68×10−8
3.20×10−8
INCREASE / (%)
+92,7
+90,4
Fig. 2. Magnetic hysteresis diagram of a typical transformer
magnetic core
159
EEEJ, Vol.1, No. 3, May 2016

During the thermal short-circuit test,
temperature increases significantly, nearly
doubling the resistance (TABLE II). The
inductance of the test object is defined by the
test objects geometric volume and the mounting
of the test loop. During testing it does not
change.
3) Circuit-breaker. A 3-phase circuit-breaker is
utilized to switch the high-current transformer
on or off. It is important to cope with the switchon and switch-off delays between the electrical
control signal and the switching operation of the
circuit breaker. The switch-off delay requires
the control system to predict the current during
the dead time. These delay times are not
constant. A statistical investigation has been
performed (Fig. 3). It shows that the standard
deviation is small compared to the mean value
of the delays (TABLE III). The switch-off
variations are inherently greater than the switchon standard deviation, because a circuit-breaker
under load current could only switch-off during
the zero crossing of the electrical current.
Due to the switch-off delay of the circuit
breaker, an extrapolation of the joule integral
shall be used to estimate the particular instant of
time at which the switch-off control command
must be issued to the circuit breaker.
To determine the optimum switch-on time, a
phase synchronous control has to be
implemented.
To determine the optimum switch-on time, the
remanence of the high-current transformer core
shall be estimated and the result be taken as set
point for synchronized switch-on command.
For the purpose of user convenience, a unique
report displaying all relevant thermal shortcircuit parameters shall be generated
automatically for each test cycle.



4. Control unit concept
A. Time-discrete calculation of the joule integral
The electrical heating of a current carrying conductor is
proportional to the applied joule integral. The joule
integral is a time-dependent function:
D. Requirements specification for the control system
t
J   i2   d 
Knowing the properties of each part of the highcurrent test circuit, the control unit has to fulfill the
following requirements.
 The discrete joule integral shall be calculated in
real-time based on the current waveform.
 For the purpose of data analysis the factual time
of open circuit transition shall be determined
and recorded.
The heating of the TO is controlled by calculating the
required Joule integral. It is defined in IEC standard
60986 [7] for aluminium conductors according to (3),
using copper conductors (4) has to be evaluated.
t
 i    d   2.19 10
2
4
2
 S  ln(
0
t
 i    d   5.11 10
2
0
S
Θsc
Θi
t
TABLE III. - Circuit-breaker delay time: statistical
distribution
ON-DELAY
86.54 ms
02.15 ms
4
2
 S  ln(
 sc  228
i  228
(3)
)
 sc  234.5
i  234.5
)
(4)
is the cross-sectional area of the
conductor in mm²
is the desired end temperature in °C
is the measured start temperature in °C
is the duration in s
For implementation in a time-discrete control platform, a
numerical integration method for determination of the
joule integral in (2) has to be selected. The accuracy of
the numerical method depends on the spacing of the time
steps and the method of numerical integration.
Basing on the sample rate of 100 kS/s of the analogdigital-converter utilized in the control system, a
comparison of different numerical integration methods
[8] has been performed for a representative current
waveform. The referenced current waveform is
analytically defined by the following formula:
Fig. 3. Distribution of the on- and off-delays of the 3-phase
circuit breaker
PARAMETER
Mean
Standard deviation
(2)
t0
i (t )  40 kA  e
OFF-DELAY
118.14 ms
008.72 ms

t
0.04 s
 30 kA  sin(2  50 Hz  t )
(5)
In TABLE IV the results of the exact analytical solution
and different numerical integration methods is presented.
160
EEEJ, Vol.1, No. 3, May 2016
TABLE IV. - Comparison of integration methods
INTEGRATION
METHOD
Analytical (exact)
Rectangle rule
Trapezoidal rule
Simpson´s rule
Simpson´s 3/8 rule
JOULE
INTEGRAL /
((kA)²s)
939.5913
939.5990
939.5910
939.5910
939.5910
RELATIVE
DIFFERENCE /
(ppm)
−
-8.12
-0.39
-0.39
-0.39
For the current waveform given in (5) the joule
integral is computed for a duration of 2 s, starting at 0 s.
With a more than sufficient accuracy and featuring low
computational efforts, the rectangle rule is selected for
implementation.
B. Prediction of open circuit transition
The switch-off delay, caused by the circuit-breaker,
makes a joule integral prediction necessary. Fig. 4 shows
the proposed method based on a linear extrapolation over
time using (6).
t
J t  t :  i
t
t0
2
  d  
tt
td
t
i
t
0
t  td
2

  d    i   d  
2
t0

Fig. 5. Joule integral prediction with different slew rate
calculations
(6)
the influence of the parameter on the quality of the
estimation results, Fig. 5 shows two different parameters.
An interval of 30 ms promises minimal dead time, but the
attenuation of the ripple is not sufficiently achieved.
Therefore, an interval of 40 ms has been selected.
This method assumes a constant root-mean-square (rms)
value of the load current during the delay time of the
system. However, the joule integral is an integration of
square of a sinusoidal function which results in the sum
of a monotonically increasing straight line and a
superimposed ripple (Fig. 5). The frequency of the ripple
is the double of the grid frequency, thus 100 Hz. If the
slew rate is calculated using an even integer multiple of
the reciprocal value (10 ms), the effects of the ripple will
be minimized.
Fig. 5 reveals this in records of actual measurements.
Using even larger integration intervals, the prediction
will become less vulnerable by noise, but firstly it will be
available at a later instant and secondly, it will react to
changes of current amplitude with a larger delay. Several
integration intervals have been tested. To demonstrate
C. Identification of closed circuit transition
The control system estimates online the joule integral.
Therefore, it is important to detect the correct start time t*
of the short-circuit test. The measured current signals are
monitored, and if the signal exceeds a threshold with
respect to interfering electromagnetic noise, the current
time is set as start time t* (Fig. 6).
D. Identification of open circuit transition
To identify the particular instant of time when the shortcircuit current is interrupted, the method as depicted
Fig. 6. Algorithm, current termination detection
Fig. 4. Algorithm, joule integral prediction
161
EEEJ, Vol.1, No. 3, May 2016
in Fig. 6 is proposed. It utilizes the insight that high
currents can only be interrupted at zero crossings. A
natural approach would be to find a zero crossing. As
there is high electromagnetic interference in the
measurement during current interruption due to arc
quenching, an additional criterion is used.
The joule integral over the last detected current half wave
is calculated and compared to a threshold. If the
calculated value is below the threshold, the end time tend
of the test is stored.
The procedure will be repeated until an open circuit
transition is recognized. Using the joule integral as
additional criterion has the advantage that high frequency
noise is attenuated, which makes the algorithm robust.
E. Phase locked loop for grid synchronous trigger
For grid synchronous triggering a phase locked loop has
been implemented using a synchronous reference frame
phase locked loop (PLL) [9]. As only one phase voltage
measurement is available and the PLL algorithm requires
a three-phase voltage measurement, a virtual 3-phase
system is created using a second order generalized
integration as proposed by Cioboaru et al. [10] (Fig. 7).
Fig. 8. Flowchart automated report generation
F. Determining the optimal phase to switch
5. Experimental results
For inductive loads optimal voltage phases are φ = 90°
and φ = 270°. Depending on the remanence of the
transformer either 90° or 270° are chosen. The
remanence can be calculated by investigating the
recorded last half-wave of the previous thermal shortcircuit test. If the current was positive in the last halfwave, the transformer should be switched-on at φ = 270°
with respect to magnetic saturation.
A. Hardware set-up and implementation
The concept introduced in section 3 has been
implemented on a rapid prototyping platform
NI compactRIO9076. For analogue data acquisition, a
NI9215 module is used with a sample rate of 100 kS/s
and a resolution of 16 bit. The short-circuit breaker is
controlled by a NI9481 relay module. The high current
transformer has a rating of 2 MVA and is capable of
short-circuit currents up to 40 kA. The current is
measured by Rogowski coil current-probe of type PEM
CWT60B for currents up to 12 kA and CWB 600B for
currents up to 120 kA. The output signal of the current
probe is connected to the analog-digital converter.
The controller includes a real-time processor and a
reconfigurable FPGA. The communication with the
controller is made through a LAN between a host PC and
the real-time operating system on the embedded
controller. All the real-time calculations and the
controlling operations of the high-current test circuit is
implemented in the FPGA.
The thermal short-circuit test cycle can be configured and
started through a graphical user interface run on the host
PC. In Fig. 9 the signal flow chart of the whole system is
shown. For reasons of storing the data in log files and
using DIADEM as PDF-creator the automated report
generation is done by the host PC.
G. Automated report generation
The automated report generation starts every time when a
test cycle has been completed. All the acquired data
stored temporarily on the control platform has to be
saved to a unique log file located on a computer at the
operator test stand.
Essential figures such as the maximum and the effective
values of the short-circuit current and the duration of the
short-circuit are computed, and visualized on a graphical
user interface.
A unique report with the measured data, the calculated
characteristic values and the general test setup
information is stored as PDF and printed out. The
flowchart for the report generation is shown in Fig. 8.
Fig. 9. Signal flow chart of the system
Fig. 7. Single-phase system phase locked loop (PLL) [12]
162
EEEJ, Vol.1, No. 3, May 2016
TABLE V. - Joule integral accuracy
PARAMETER
Average
Standard deviation
With correct phase setting, no voltage sag in the recorded
network voltage is observed, and the TO voltage is
sinusoidal. With incorrect phase setting, a network
voltage sag of 6.5 % occurs and the voltage on the TO
harmonic distortion due to magnetic core saturation.
DIFFERENCE / (%)
0.20
1.96
B. Determination of control system performance
For first functional checks of the control unit, an air coil
with an inductance of 2 mH has been connected as test
object. The high-current transformer is configured to a
secondary voltage of 173 V.
To determine the joule integral accuracy, 26
measurements with different joule integral set points
have been executed. TABLE V shows the statistical
characteristics of the differences to the set point.
Fig. 10 shows exemplary the test object current and the
joule integral estimated by the real-time controller. The
set point was 0.04 kA²s. When the predicted joule
integral is reached at approx. 460 ms, the switch off
command is issued (section 3.2) to the circuit-breaker.
Fig. 11 shows the different behavior of the current when
using different phase angles to switch-on the transformer.
An optimal switch-on phase of φ = 90° is recorded in the
upper graph. For comparison, a short-circuit test with a
switch-on phase of φ = 0° has been recorded in the lower
graph of Fig. 11. An excessively large inrush current
superimposes the steady-state short-circuit current.
In Fig. 12 the influence of the remanence is shown. Phase
angle was set in both test to the same value of φ = 90°.
The only difference is that in the upper diagram, the
transformer had a positive remanence and in the lower
diagram the remanence was negative.
Fig. 11. Measurement air coil φ = 90°, φ = 0°
Fig. 10. Exemplary short-circuit test data
Fig. 12. Experimental test of influence of transformer
remanence
163
EEEJ, Vol.1, No. 3, May 2016
C. Thermal short-circuit testing
5. Conclusion
For final commissioning, a 1,000 mm² copper rail has
been connected as test object to the high-current
transformer, which is configured for a secondary voltage
of 70 V. With an calculated resistance of 6.1 µΩ and an
inductance of 6.5 µH, a steady stead current of approx.
20 kA is expected. The control unit is configured to a
joule integral setpoint of 300.0 (kA)²s.
Fig. 13 shows the recorded values, and the evaluated
parameters of the test are given in TABLE VI. As it can
be seen, the control unit performs the high-current tests
with a high accuracy.
Finally in Fig. 14 an automatically generated report is
reproduced. As TO a 300 mm² aluminium conductor was
used. The temperature set point was 255 °C. To reach
this end temperature, a joule integral set point of
1313 (kA)²s was calculated using equation (3.2). A joule
integral of 1316 (kA)²s has been achieved, resulting in an
end temperature of 255.5 °C.
Conductors with high cross-sectional areas are nowadays
widely introduced for network connection of renewable
electricity generation. These components are subjected to
standardized testing [6]. Increasing cross-sectional areas
require higher rated and more accurate controlled thermal
short-circuit tests stands.
A real-time control of the short-circuit current duration is
proposed in this paper. By determining the specific
energy the test object is subjected to, it effectively deals
with nonlinear decreasing test currents. Dead times for
short-circuit interruption are considered using a
prediction of the measured current values. In addition, the
control system features a phase locked loop for network
voltage synchronization, allowing an optimized switchon phase for minimized network disturbance during
testing. For convenience, the control systems assists the
test stand operator with an automated evaluation of each
thermal short-circuit.
The control concept has been implemented on a rapid
prototyping solution and used for experimental testing on
a high-current transformer. The results prove reliable
operation of the test system and a high accuracy of the
control solution, fulfilling standardized requirements.
TABLE VI. - Results of 1,000 mm² copper rail measurement
PARAMETER
Secondary voltage
I²t set point
I²t achieved
Ipeak
Irms
Duration t
VALUE
70 V
300.0 (kA)²s
303.0 (kA)²s
38.6 kA
22.7 kA
588.3 ms
References
[1] Network transparency internet information platform of the
German transmission system operators,
http://www.netztransparenz.de, last accessed 2015-06-10.
[2] Z. Chen, “Issues of Connecting Wind Farms into Power
Systems”, in Proc. of the 2005 IEEE/PES Transmission
and Distribution Conference & Exhibition: Asia and
Pacific, Vol. 1, pp. 1-6, Dalian, China, 2005.
[3] German association of energy and water industries
(BDEW), “Technical Guideline: Generating Plants
Connected to the Medium-Voltage Network”, 2008.
[4] IEC international standard 61238-1:2003, “Compression
and mechanical connectors for power cables for rated
voltages up to 30 kV (Um = 36 kV)”, 2003.
[5] IEC international standard 60502-4:2010, “Power cables
with extruded insulation and their accessories for rated
voltages from 1 kV (Um = 1,2 kV) up to 30 kV (Um =
36 kV) - Part 4: Test requirements on accessories for
cables with rated voltages from 6 kV (Um = 7,2 kV) up to
30 kV (Um = 36 kV)”, 2010.
[6] IEC international standard 61442:2005, “Test methods for
accessories for power cables with rated voltages from
6 kV (Um = 7,2 kV) up to 30 kV (Um = 36 kV)”, 2005.
[7] IEC international standard 60986:2000, “Short-circuit
temperature limits of electric cables with rated voltages
from 6 kV (Um = 7,2 kV) up to 30 kV (Um = 36 kV)”,
2000.
[8] A.R. Krommer and C.W. Ueberhuber, “Numerical
Integration: on Advanced Computer Systems”, Springer,
Berlin Heidelberg, 1994
[9] L.N Arruda, S.M. Silva and B.J.C Filho, “PLL structures
for utility connected systems,” Conference Record of the
2001 IEEE Industry Applications Conference, Vol. 4, pp.
2655-2660, Chicago, USA, 2001.
[10] M. Ciobotaru, R. Teodorescu, and F. Blaabjerg, “A New
Single-Phase PLL Structure Based on Second Order
Generalized Integrator”, in Proc. of the 37th IEEE Power
Electronics Specialists Conference (PESC ’06), pp. 1-6,
Orlando, USA, 2006.
Fig. 13. Oscillogram of final commissioning
CH1: Test object current, 20 kA/div (blue)
CH2: Reference network voltage, 500 V/div (cyan)
CH4: Test object voltage, 100 V/div (green)
Time scale: 100 ms/div
Fig. 14. Example of an automatically generated report, test
object: 300 mm² aluminium connector
164
EEEJ, Vol.1, No. 3, May 2016